This is, to our knowledge, the first study on CVRM in European primary care at a larger scale. We found that scores on quality indicators in general vary from 45% (a record of advice on physical activity; SBP and LDL below treatment targets) up to about 95% (blood pressure recording) of the maximum score, which indicates optimal policy. As opposed to our expectation, we found little evidence for better performance in large practices. In Slovenia larger practices tended to perform better. Our study did not explicitly assess the efficiency of delivering CVRM.
Similar to our research, the three EUROASPIRE surveys provide data from international research with uniform data collection across countries. But in EUROASPIRE a specialist care starting point guided CHD patient selection . Raised blood pressure, defined as SBP≥140 mmHg and/or DBP≥90 mmHg or in diabetics respectively ≥130 mmHg and ≥80 mmHg, was prevalent in 58-61% of these survey samples. Raised cholesterol was defined as ≥4.5 mmol/l and diminished from 94.5% in EUROASPIRE I, to 76.7% in the second survey and finally to 46.2% in the third. Data collected in the most recent EUROASPIRE survey in 2006 and 2007 are comparable to our results. In the Pinnacle programme, data regarding outpatients from cardiology offices, too, show comparable results with for instance antiplatelet and statin therapy in 84.9% and 84.3%, respectively . Previous data on CVRM in primary care can be found in various national studies. In a Cochrane review the effects of interventions on the organisation of the treatment considering ischemic heart disease patients in primary care are studied. . Data from the control groups could be considered comparable to our audit data. Direct comparable outcomes are statin prescription and antiplatelet therapy. In the review 50.1 of the control patients received statin therapy, but studies dated back till the 1990’s. The most recent study, SPHERE, had with 80.3% a result comparable to 82.7% in our study. Relating to antiplatelet therapy the review result was 72.5% compared to 87.7 in our study sample. Again, the more recent data were the best, up to 87.0 In the SHERE study SBP was <140 mmHg in 66.2% (versus 47.5% in our data) and DBP<90 mmHg in 88.6% of the patients (comparable to 85.5% in our data). In drug trials efficacy of statins varies from 60 to 90% in achieving LDL < 2.5 mmol/l [26–29]. In our observational study, the real life results are on the lower end of this range. In contrast to the optimum situation in these drug studies physicians could include every patient known to have a CHD, patients without further medical attention, too. The indicator on LDL cholesterol treatment target surprisingly was not validated in the Delphi indicator development procedure. We can only speculate about the reasons; setting strict norms irrespective of the patient’s age might be argued by some or the fact that this outcome measure very much depends on the patient in contrast to process measures as offering a statin. In view of the strong evidence base for the relationship between LDL cholesterol and coronary heart disease we anyhow decided to include the LDL cholesterol results in our study.
Since most patients with increased cardiovascular risk are treated in primary care, the findings are relevant for improving care in the different countries despite study limitations. They show that specific countries scored high on some indicators and low on others. Improvements in CVRM are possible in all countries. Our study allowed to include all patients with a known diagnosis of CHD. Inevitably, patients treated in secondary care could be included, too. Our results give an overview of the performance of CVRM related to all patients known in the primary care practice.
In England high scores on performance indicators were observed, particularly for indicators incentivized as part of the Quality and Outcomes Framework (QOF) . Physicians in England are forced, by their electronic patient records, to tick boxes for QOF-indicators, which might be a strong driver for change in registration, enhancing good risk factor registration in England. On the other hand, we found relatively low performance scores for some indicators of CVRM, especially risk factor recording, in the Netherlands. The only indicator related to a financial incentive (influenza vaccination) and supported by a national organizational programme had very high scores in this country (in 2009 a fee of 9.88 euro was provided for every vaccinated patient). The system parameter incentives on a national level and as such as a country characteristic may have an important influence relative to practice size as a practice characteristic.
The DBP indicator scores were much higher than scores for SBP, though the importance of the latter is stressed by its role in risk classification schemes. Advice on physical activity had low scores, too, although it remains uncertain whether such advice had been provided but not recorded.
Differences between countries may be partly explained by differences in the quality of recording as stated above. Medication and blood pressure or cholesterol levels are probably well recorded, but this is less the case for smoking status or exercise advice. It might be argued that recorded care does not mirror care provided. But in chronic care recording is thought to be essential. Risk factor recording is a prerequisite to select patients for treatment and chronic care means collaboration between various health care professionals, who will need to rely on the data in the patient records .
In our study practice size seems to have little relation to performance as measured by quality indicators. Though in previous research on practice size no consistent results were found, in general larger practices tend to show better performances and provide more extensive services, for instance more preventive activities [1–5]. All these studies were based on national data. We took into account the fact that we had practices from eight countries by entering country as a level in our multilevel analysis. This procedure effected chance on significant findings. Taking into account the strength of the primary care did not provide relevant findings.
A larger practice offers opportunities to develop skills by experience and gives managerial advantages, especially when specialized staff is required. Structured care will be more cost effective with larger patient groups included in a program. On the other hand, there seems to be a trade of between high quality clinical care and interpersonal care, and access might be better in smaller practices [2, 9]. In our sample across countries small practices were able to deliver a performance on cardiovascular risk management as good as larger practices.
Only in Slovenia larger practices showed a tendency towards better performance in general. We can only hypothize on this finding. It may be the resultant of recent implementation strategies with first effects in larger settings. This would be in line with the general concept of larger practices being in a good position for providing structured care to larger groups of patients.
The proportion of variance explained at the practice level was larger than that related to the country level, indicating that the practice has more influence on that variation than the country. This could stimulate practices to invest in quality improvement in their practices as there is little argument that much is determined at a higher level out of their reach. A remarkable small part of the variation in outcomes is explained at both the practice and the country level considering the blood pressure and cholesterol levels. These biological outcomes will be determined at the patient level to a greater extend.
Strengths and limitations of the study
Within the context of our international survey we had to face inclusion bias both at the practice level and at the patient level as a result of differences in the organization of the health care system within the various countries at different levels. Practice selection was random in most countries but a convenience sample in two countries (Austria and Switzerland). The procedure for sampling patients, too, showed some variation. In Belgium, England, the Netherlands and Slovenia patient selection was exclusively based on recorded diagnoses, enabling inclusion of patients registered but not controlled in primary care or not at all. Less strict methods were used in the other countries (remembering patients, prescription lists, attending patients) providing patient inclusion bias. Patients on a prescription list by definition have some drug treatment and frequent attenders and treated patients are more likely to be remembered. Our practice sample appeared the best feasible given the limitations of our international survey. The sample size of 181 practices forms a limitation to detect small effects of practice size on outcomes, among other due to clustering within countries and differences of possible effects between countries.
Data on practice size were not directly comparable between countries because of the differences in health care systems. In some but not all countries patients are listed with one GP or practice. Countries without these clear patients’ lists had to report on numbers of attenders as a measure for practices size. By standardizing practice size data per country we solved this potential problem.
We included patients with CHD to have a patient group more homogeneous than the group of CVD patients in general. This did not completely prevent heterogeneity within our study population. The CHD group comprised on the one hand patients who had a myocardial infarction or vascular surgery and have been treated in secondary care and on the other hand patients with stable angina pectoris who might have been treated in primary care exclusively.
The variation between practices within each country is unwanted and proves potential for improvement. The presence of highly performing practices within each country proves that in each national context good CVRM is possible. Differences found between countries and especially best practices can form lessons for all countries. For instance the Quality and Outcomes Framework from the UK can be an example to other countries but focus may differ according to the national situation as the position of primary care within the larger context of the health care system.
In contrast to most previous research our analysis did not indicate significant influence of practice size on the quality indicator scores. In various studies larger practices tend to perform better, supporting the development of practice collaboration with consequently larger groups of CHD patients to organize care. This may enhance expertise and logistics. We could not confirm this tendency. Here, further research is needed.